Detecting Holes in Fish Farming nets: A Two–Method approach

This paper presents two different techniques for detecting holes in fish farming nets using image processing. The first technique utilizes the OpenCV library, while the second employs artificial neural networks (ANNs). A real-world dataset of fish farming net images, captured by an ROV with various hole types, was used to evaluate the performance of both techniques. The results indicate that the OpenCV-based approach outperforms the ANN-based method in terms of accuracy and robustness, despite the fact that both methods performed poorly. However, additional research can significantly improve the proposed techniques, ultimately enhancing the efficiency and reliability of hole detection in fish farming nets. This is particularly important for maintaining the health and productivity of fish in aquaculture operations.